Upload
anne-campbell
View
213
Download
1
Embed Size (px)
Citation preview
Discussion of„Employment risk and the
living arrangements of young adults“
Martin Biewen, University of Mainz
Summary
Does increased employment risk make young Spaniards stay longer with their parents?
Idea: measure employment risk by type of labour contract (temporary vs. permanent)
Problem: potential endogeneity of contract type Solution: seek exogenuous variation driving
contract type: 1) legal time limits 2) subsidization
IV-estimates: no or only weak causal effects of employment risk on decision to leave nest
Comments: 1) General
Relevant and interesting question Paper makes serious effort to identify causal
effect Use of different samples and models Results interesting because they are
unexpected
Comments: 2) LATE
IV-estimate is local average treatment effect Effect only refers to individuals who change contract
type as a consequence of changes of the IV These individuals are not necessarily representative Legal time limits: those individuals who don‘t find
permanent employment before 3 years limit (the less able, less motivated, less emancipated?)
Subsidization: those individuals who wouldn‘t have found permanent employment if subsidization had been lower (the less able, less motivated, less emancipated?)
Comments: 3) Flows approach
Flows approach (= quarterly transitions) doesn‘t seem to contain much information
Incidence of contract changes in sample of individuals with temporary contracts only 2.2 % (! tiny „treatment“ group)
Only a fraction of these will change type due to IV Looking at the coincidence of contract change and
moving out in the same quarter seems restrictive Flows sample only looks at individuals who start
with temporary contract (representative?, efficiency?)
! stocks approach much better
Comments: 4) Data limitations
Provide more info on data in paper Rotating Panel: each individual observed for
„five quarters“ only (p. 13)? In addition: attrition?
Potential Tenure variable: Biographic info? Yearly measure? Sequences of temporary jobs?
Subsidies variable: firm specific? (if so: potentially correlated with unobserved individual characteristics)
Provide more info on incidence of different tempo-rary contracts (one type of contract has a 2 years time limit and one type disappears in´97)
Comments: 5) Specific points If 70% of individuals between 18 and 35 year still
live with parents ! extend age range to observe when individuals actually move out
Age is important variable ! include not only linearly Regressors „Exp“ and „Expc“ (= Exp -3) unclear
(negative values, collinearity?) Why no „stocks regressions“ in subsidies case? Aggregate employment risk: regional unemploy-
ment rate missing in time limits regressions! Better include age-specific unemployment rates
Comments: 5) Specific points
Difference in „employment risk“ between tem-porary and permanent jobs large enough to drive decisions?
12.5 %
2.5 %
Comments: 6) Alternative models
So far, paper doesn‘t exploit information about employment status
Include employment risk explicitly as regressor
„Employment risk“, e.g. Probit of employment status on exogenous variables and instruments
is uncorrelatedwith if and are exogenous
Comments: 7) Alternative models
Timing-of-Events approach (Abbring/V.d.Berg (2003))
No exclusion restrictions needed But: potential problem with anticipation of
conversion
Hazard of conversion to permanent contract
Hazard of leaving parents
unobserved heterogeneity
potential jump in hazard after conversion
Comments: 8) Other explanations
Cultural differences (role of family)/cultural change
Longer education Women‘s emancipation/women‘s
participation in higher education Parents richer than in the past Underdevelopped housing markets (rented
accomodation)? … ?